Current Journal of Applied Science and Technology

Time series modelling has been proved its usefulness in various fields including meteorology, hydrology and agriculture. It utilizes past data and extracts useful information from them to build up a model which could simulate various processes. The prior knowledge of evapotranspiration could help in estimating the amount of water required by the crops that is useful for optimizing design of irrigation systems. In this study, the time series modelling of monthly temperature and reference evapotranspiration has been carried out utilizing past data of 35 years (1983-2017) to assist decision makers related to agriculture and meteorology. 30 years (1983-2012) of temperature and evapotranspiration data were used for training and remaining 5 years of data (2013-2017) were used for validation. The monthly evapotranspiration was estimated using Penman-Monteith FAO-56 method. Mann-Kendall test was used at 5% significant level for identifying trend component in mean temperature. The time series of temperature and evapotranspiration was made stationary for modelling the stochastic components using ARIMA (Autoregressive Integrated Moving Average) model. In order to check the normality of residuals, the Portmantaeu test was applied. The time series models for temperature and evapotranspiration which were validated for 5 years (2013-2017) and further deployed for forecasting of 5 years (2018-2022). It was found that for modelling temperature and reference evapotranspiration for Navsari, seasonal ARIMA (1,0,0)(0,1,1)12 and seasonal ARIMA (1,0,1)(1,1,2)12 were found to be appropriate models respectively. Mann Kendall test used for trend detection in monthly mean temperature revealed that October and November months had significant positive trend. Negative trend was observed only in the month of June.